xen discuss - Aug 2006 - ethernet MAC addresses for Solaris domU

If I understand things correctly each Solaris domU will need to have its 
own ethernet address assigned to it.  Do we just "make these up" or is
there a pool that should be getting used ?

I attempted to create a Solaris domU without specifying an ethernet 
address and I got the following error message when running ''xm
create''

# vbdcfg mkdomU -f fileproto wallace
# xm create -c wallace-64.py
Using config file "wallace-64.py".
Error: Device 0 (vif) could not be connected. Hotplug scripts not working.
# cat wallace-64.py
name = "wallace"
memory = "400"
kernel = "/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix"
extra = "/platform/i86xen/kernel/amd64/unix"
ramdisk = "/export/xc/xvm/wallace/platform/i86pc/boot_archive"
on_shutdown = "destroy"
on_reboot = "restart"
on_crash = "preserve"
root = "/dev/vdsk/d0s0"
disk = [''file:/export/xc/xvm/wallace/root.file,0,w'']
vif = [''bridge=xenbr0'']

Is the fact that I didn''t specify an ethernet address the problem ?

-- 
Darren J Moffat

Darren J Moffat wrote:

 >>Is the fact that I didn''t specify an ethernet address the
problem ?

I don''t think so. 

Check the xen Logfiles (/var/log/xen*log) - if you can''t find anything
there
try starting the DomU again after rebooting the Dom0



regards

Bernd

>
> I attempted to create a Solaris domU without specifying an ethernet 
> address and I got the following error message when running ''xm
create''
>
> # vbdcfg mkdomU -f fileproto wallace
> # xm create -c wallace-64.py
> Using config file "wallace-64.py".
> Error: Device 0 (vif) could not be connected. Hotplug scripts not 
> working.
> # cat wallace-64.py
> name = "wallace"
> memory = "400"
> kernel =
"/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix"
> extra = "/platform/i86xen/kernel/amd64/unix"
> ramdisk = "/export/xc/xvm/wallace/platform/i86pc/boot_archive"
> on_shutdown = "destroy"
> on_reboot = "restart"
> on_crash = "preserve"
> root = "/dev/vdsk/d0s0"
> disk = [''file:/export/xc/xvm/wallace/root.file,0,w'']
> vif = [''bridge=xenbr0'']
>
> Is the fact that I didn''t specify an ethernet address the problem
?
>

-- 
Bernd Schemmer, Frankfurt am Main, Germany
http://home.arcor.de/bnsmb/index.html

Más temprano que tarde el mundio cambiará
                        Fidel Castro

Bernd Schemmer wrote:
> Darren J Moffat wrote:
> 
>  >>Is the fact that I didn''t specify an ethernet address the
problem ?
> 
> I don''t think so.
Correct, I have many domUs without mac entries...

> Check the xen Logfiles (/var/log/xen*log) - if you can''t find
anything
> there
> try starting the DomU again after rebooting the Dom0
look at /var/log/xend.log.  Start at the end and work backwards.
It''s very wordy so it''s pretty easy to see where you started
the domU and where it shutdowns.

There is a bug currently, when you specify a bad disk path
in the config file, the disk backend fails non gracefully.
This sounds like a similar type of bug, but the the network
backend. It would be very useful yo track this down so we could
fix it..

If you did restart the dom0 and it works, please keep an eye
out for the failure..


Thanks,

MRJ



> 
> 
> regards
> 
> Bernd
> 
> 
>>
>> I attempted to create a Solaris domU without specifying an ethernet 
>> address and I got the following error message when running ''xm
create''
>>
>> # vbdcfg mkdomU -f fileproto wallace
>> # xm create -c wallace-64.py
>> Using config file "wallace-64.py".
>> Error: Device 0 (vif) could not be connected. Hotplug scripts not 
>> working.
>> # cat wallace-64.py
>> name = "wallace"
>> memory = "400"
>> kernel =
"/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix"
>> extra = "/platform/i86xen/kernel/amd64/unix"
>> ramdisk =
"/export/xc/xvm/wallace/platform/i86pc/boot_archive"
>> on_shutdown = "destroy"
>> on_reboot = "restart"
>> on_crash = "preserve"
>> root = "/dev/vdsk/d0s0"
>> disk = [''file:/export/xc/xvm/wallace/root.file,0,w'']
>> vif = [''bridge=xenbr0'']
>>
>> Is the fact that I didn''t specify an ethernet address the
problem ?
>>
> 
>

Darren J Moffat wrote:
> If I understand things correctly each Solaris domU will need to have its 
> own ethernet address assigned to it.  Do we just "make these up"
or is
> there a pool that should be getting used ?
No pool currently.. As far as what we will be doing, dme
should have a much better idea...

> Is the fact that I didn''t specify an ethernet address the problem
?
If you don''t specify a MAC, xen will make up one semi-randomly.
This is a temporary solution.



MRJ

Mark Johnson wrote:
> look at /var/log/xend.log.  Start at the end and work backwards.
> It''s very wordy so it''s pretty easy to see where you
started
> the domU and where it shutdowns.
> 
> There is a bug currently, when you specify a bad disk path
> in the config file, the disk backend fails non gracefully.
> This sounds like a similar type of bug, but the the network
> backend. It would be very useful yo track this down so we could
> fix it..
> 
> If you did restart the dom0 and it works, please keep an eye
> out for the failure..
I restarted the dom0 and it still fails.  The full xend.log file
from the attempt at xm create is attached.

What concernes me is the number of references to /local and to linux
neither of which I have this is a Solaris domU on a Solaris dom0.

-- 
Darren J Moffat

Darren J Moffat wrote:
> Mark Johnson wrote:
>> look at /var/log/xend.log.  Start at the end and work backwards.
>> It''s very wordy so it''s pretty easy to see where you
started
>> the domU and where it shutdowns.
>>
>> There is a bug currently, when you specify a bad disk path
>> in the config file, the disk backend fails non gracefully.
>> This sounds like a similar type of bug, but the the network
>> backend. It would be very useful yo track this down so we could
>> fix it..
>>
>> If you did restart the dom0 and it works, please keep an eye
>> out for the failure..
> 
> I restarted the dom0 and it still fails.  The full xend.log file
> from the attempt at xm create is attached.
> 
> What concernes me is the number of references to /local and to linux
> neither of which I have this is a Solaris domU on a Solaris dom0.

The /local references are virtual paths into the xen "database"
called xenstore.. You can do things like xenstore-ls to examine it.
(not that that would help you here, just fyi)

The Linux references are just "prints" which need to be cleaned up
in the xen domain builder code.. There''s a lot of code there that
needs to be cleaned up.

Once your domain fails to start like this, the disk backend is done
(again, known bug being worked on right now) so all further attempts
to start a domU (with a disk) are futile until you reboot the dom0.

Looking at this and your previous post, your root path may be incorrect?
You have
    root=/dev/vdsk/d0s0

Which is the way it used to be. But I had thought we fixed before
the last release (e.g. now /dev/dsk/c0d0s0). I could be wrong though.
Here''s is my conf file (I''ve modified the paths to my liking).

I''ll send you a link to some pre-configured domUs offline
and we''ll see if those work..

  solaris-11[1]#; cat 32-solaris.py
name = "solaris"
memory = "384"
kernel =
"/export/xc/xvm/kernels/solaris-snv_43/platform/i86xen/kernel/unix"
extra = "/platform/i86xen/kernel/unix"
ramdisk =
"/export/xc/xvm/kernels/solaris-snv_43/platform/i86pc/boot_archive"
on_shutdown = "destroy"
on_reboot = "restart"
on_crash = "destroy"
root = "/dev/dsk/c0d0s0"
disk = [''file:/export/xc/xvm/disks/solaris-snv_43.img,0,w'']
vif = [''bridge=xenbr0'']



Thanks,


MRJ

> 
> ------------------------------------------------------------------------
> 
> [2006-08-11 05:58:47 xend.XendDomainInfo] DEBUG (XendDomainInfo:178)
XendDomainInfo.create([''vm'', [''name'',
''wallace''], [''memory'',
''400''], [''on_reboot'',
''restart''], [''on_crash'',
''preserve''], [''vcpus'', 1],
[''image'', [''linux'',
[''kernel'',
''/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix''],
[''ramdisk'',
''/export/xc/xvm/wallace/platform/i86pc/boot_archive''],
[''root'', ''/dev/vdsk/d0s0''],
[''args'',
''/platform/i86xen/kernel/amd64/unix'']]],
[''device'', [''vbd'',
[''uname'',
''file:/export/xc/xvm/wallace/root.file''],
[''dev'', ''0''], [''mode'',
''w'']]], [''device'', [''vif'',
[''bridge'', ''xenbr0'']]]])
> [2006-08-11 05:58:47 xend.XendDomainInfo] DEBUG (XendDomainInfo:284)
parseConfig: config is [''vm'', [''name'',
''wallace''], [''memory'',
''400''], [''on_reboot'',
''restart''], [''on_crash'',
''preserve''], [''vcpus'', 1],
[''image'', [''linux'',
[''kernel'',
''/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix''],
[''ramdisk'',
''/export/xc/xvm/wallace/platform/i86pc/boot_archive''],
[''root'', ''/dev/vdsk/d0s0''],
[''args'',
''/platform/i86xen/kernel/amd64/unix'']]],
[''device'', [''vbd'',
[''uname'',
''file:/export/xc/xvm/wallace/root.file''],
[''dev'', ''0''], [''mode'',
''w'']]], [''device'', [''vif'',
[''bridge'', ''xenbr0'']]]]
> [2006-08-11 05:58:47 xend.XendDomainInfo] DEBUG (XendDomainInfo:380)
parseConfig: result is {''ssidref'': None,
''uuid'': None, ''on_crash'':
''preserve'', ''on_reboot'':
''restart'', ''image'':
[''linux'', [''kernel'',
''/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix''],
[''ramdisk'',
''/export/xc/xvm/wallace/platform/i86pc/boot_archive''],
[''root'', ''/dev/vdsk/d0s0''],
[''args'',
''/platform/i86xen/kernel/amd64/unix'']],
''on_poweroff'': None, ''cpus'': None,
''name'': ''wallace'',
''backend'': [], ''vcpus'': 1,
''cpu_weight'': None, ''vcpu_avail'': None,
''memory'': 400, ''device'':
[(''vbd'', [''vbd'', [''uname'',
''file:/export/xc/xvm/wallace/root.file''],
[''dev'', ''0''], [''mode'',
''w'']]), (''vif'', [''vif'',
[''bridge'', ''xenbr0'']])],
''bootloader'': None, ''cpu'': None,
''maxmem'': None}
> [2006-08-11 05:58:47 xend.XendDomainInfo] DEBUG (XendDomainInfo:1164)
XendDomainInfo.construct: None 0
> [2006-08-11 05:58:47 xend.XendDomainInfo] DEBUG (XendDomainInfo:1196)
XendDomainInfo.initDomain: 3 1.0
> [2006-08-11 05:58:47 xend] DEBUG (balloon:126) Balloon: free 1496; need
401; done.
> [2006-08-11 05:58:47 xend] INFO (image:135) buildDomain os=linux dom=3
vcpus=1
> [2006-08-11 05:58:47 xend] DEBUG (image:173) dom            = 3
> [2006-08-11 05:58:47 xend] DEBUG (image:174) image          =
/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix
> [2006-08-11 05:58:47 xend] DEBUG (image:175) store_evtchn   = 1
> [2006-08-11 05:58:47 xend] DEBUG (image:176) console_evtchn = 2
> [2006-08-11 05:58:47 xend] DEBUG (image:177) cmdline        = 
root=/dev/vdsk/d0s0 /platform/i86xen/kernel/amd64/unix
> [2006-08-11 05:58:47 xend] DEBUG (image:178) ramdisk        =
/export/xc/xvm/wallace/platform/i86pc/boot_archive
> [2006-08-11 05:58:47 xend] DEBUG (image:179) vcpus          = 1
> [2006-08-11 05:58:47 xend] DEBUG (image:180) features       = 
> [2006-08-11 05:58:50 xend] DEBUG (blkif:24) exception looking up device
number for 0: [Errno 2] No such file or directory: ''/dev/0''
> [2006-08-11 05:58:50 xend] DEBUG (DevController:103) DevController: writing
{''virtual-device'': ''0'',
''backend-id'': ''0'',
''state'': ''1'', ''backend'':
''/local/domain/0/backend/vbd/3/0''} to
/local/domain/3/device/vbd/0.
> [2006-08-11 05:58:50 xend] DEBUG (DevController:105) DevController: writing
{''domain'': ''wallace'',
''frontend'': ''/local/domain/3/device/vbd/0'',
''dev'': ''0'', ''state'':
''1'', ''params'':
''/export/xc/xvm/wallace/root.file'', ''mode'':
''w'', ''frontend-id'': ''3'',
''type'': ''file''} to
/local/domain/0/backend/vbd/3/0.
> [2006-08-11 05:58:50 xend] DEBUG (DevController:103) DevController: writing
{''backend-id'': ''0'', ''mac'':
''00:16:3e:39:be:d2'', ''handle'':
''0'', ''state'': ''1'',
''backend'':
''/local/domain/0/backend/vif/3/0''} to
/local/domain/3/device/vif/0.
> [2006-08-11 05:58:50 xend] DEBUG (DevController:105) DevController: writing
{''bridge'': ''xenbr0'',
''domain'': ''wallace'',
''handle'': ''0'', ''script'':
''/etc/xen/scripts/vif-bridge'', ''state'':
''1'', ''frontend'':
''/local/domain/3/device/vif/0'', ''mac'':
''00:16:3e:39:be:d2'', ''frontend-id'':
''3''} to /local/domain/0/backend/vif/3/0.
> [2006-08-11 05:58:50 xend.XendDomainInfo] DEBUG (XendDomainInfo:671)
Storing VM details: {''ssidref'': ''0'',
''uuid'':
''9d2e7f92-dddd-099c-fab2-4846aa57c85a'',
''on_reboot'': ''restart'',
''start_time'': ''1155272330.22'',
''on_poweroff'': ''destroy'',
''name'': ''wallace'',
''vcpus'': ''1'',
''vcpu_avail'': ''1'',
''memory'': ''400'',
''on_crash'': ''preserve'',
''image'': ''(linux (kernel
/export/xc/xvm/wallace/platform/i86xen/kernel/amd64/unix) (ramdisk
/export/xc/xvm/wallace/platform/i86pc/boot_archive) (root /dev/vdsk/d0s0) (args
/platform/i86xen/kernel/amd64/unix))'', ''maxmem'':
''400''}
> [2006-08-11 05:58:50 xend.XendDomainInfo] DEBUG (XendDomainInfo:696)
Storing domain details: {''console/ring-ref'':
''95425'', ''console/port'':
''2'', ''name'': ''wallace'',
''console/limit'': ''1048576'',
''vm'':
''/vm/9d2e7f92-dddd-099c-fab2-4846aa57c85a'',
''domid'': ''3'',
''cpu/0/availability'': ''online'',
''memory/target'': ''409600'',
''store/ring-ref'': ''95426'',
''store/port'': ''1''}
> [2006-08-11 05:58:50 xend] DEBUG (DevController:132) Waiting for devices
vif.
> [2006-08-11 05:58:50 xend.XendDomainInfo] DEBUG (XendDomainInfo:877)
XendDomainInfo.handleShutdownWatch
> [2006-08-11 05:58:50 xend] DEBUG (DevController:138) Waiting for 0.
> [2006-08-11 05:58:50 xend] DEBUG (DevController:403) hotplugStatusCallback
/local/domain/0/backend/vif/3/0/hotplug-status.
> [2006-08-11 05:59:00 xend.XendDomainInfo] DEBUG (XendDomainInfo:1327)
XendDomainInfo.destroy: domid=3
> [2006-08-11 05:59:00 xend.XendDomainInfo] DEBUG (XendDomainInfo:1335)
XendDomainInfo.destroyDomain(3)
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> xen-discuss mailing list
> xen-discuss@opensolaris.org

Mark Johnson wrote:
> Looking at this and your previous post, your root path may be incorrect?
> You have
>    root=/dev/vdsk/d0s0
> 
> Which is the way it used to be. But I had thought we fixed before
> the last release (e.g. now /dev/dsk/c0d0s0). I could be wrong though.
> Here''s is my conf file (I''ve modified the paths to my
liking).
That was the problem, but the error message was a but vif so I hadn''t 
been looking at disk things.

This was caused by me using a version of vdbcfg that was not the one in 
the current SUNWxen package - operator error!

Thanks my domU is now up and running.

I can now start figuring out how things work so I can start looking at 
how to give access to crypto hardware to a Solaris domU from a Solaris 
dom0 - we have a prototype for sun4v LDoms and I''d like to try and
reuse
as much of that as possible.

-- 
Darren J Moffat